Thursday, April 21st, 2011 at 4.10pm, MSB 1147 (Colloquium Room)
Refreshments: 3.30pm, MSB 4110 (Statistics Lounge)
Speaker: Andrew Latimer (Dept. of Plant Sciences, UC Davis)
Title: Hierarchical Models for Plant Species Distributions
Abstract: Ecologists often want to use patterns of observed abundances of species to infer their relationships with environmental factors such as temperature and soil type, and to predict their probability of presence or expected abundances in new locations or under changed conditions. Standard approaches to such "species distribution models" relate observations of presence/absence or abundance directly to environmental variables to draw these inferences. In some cases, however, this direct approach can be misleading because other factors such as land use can have substantial effects on observed species distributions. This talk describes statistical models of plant species distributions that use a hierarchical Bayesian approach to take into account two important factors: land use transformation and observer error. Models are applied to extensive presence/absence and ordinal abundance data for South African shrub species in family Proteaceae in the Cape region of South Africa. We use the models to reconstruct "potential" distribution and abundance of species, and to infer the importance of land use and environmental variables for the distribution of these species. Related ongoing work to model frequency of wildfires and rates of vegetation regrowth after fire will also be briefly described.
